src.dackar.RCA.orchestrators.epistemics_digest
==============================================

.. py:module:: src.dackar.RCA.orchestrators.epistemics_digest


Attributes
----------

.. autoapisummary::

   src.dackar.RCA.orchestrators.epistemics_digest.JsonDict


Functions
---------

.. autoapisummary::

   src.dackar.RCA.orchestrators.epistemics_digest._authority_level
   src.dackar.RCA.orchestrators.epistemics_digest._classification_resolution_level
   src.dackar.RCA.orchestrators.epistemics_digest._is_degraded
   src.dackar.RCA.orchestrators.epistemics_digest.build_epistemics_digests
   src.dackar.RCA.orchestrators.epistemics_digest.build_epistemics_run_summary


Module Contents
---------------

.. py:data:: JsonDict

.. py:function:: _authority_level(meta)

   Map authority rank to §7.4 level string.


.. py:function:: _classification_resolution_level(meta)

   How the epistemic class was determined for this hit.

   Prefers the explicit ``classification_resolution_level`` written upstream by
   ``doc_extraction`` (``finding_status`` / ``authority_level`` / ``doc_type`` /
   ``default``); only when that provenance is absent does it fall back to
   inferring the level locally.


.. py:function:: _is_degraded(meta)

   Whether this hit's epistemic classification is degraded.

   Prefers the explicit ``degraded_classification`` flag written upstream by
   ``doc_extraction``; only when that flag is absent does it fall back to
   treating a hit with no ``epistemic_class`` annotation as degraded.


.. py:function:: build_epistemics_digests(causality_candidates, results)

   Return {candidate_id → EpistemicsDigest dict} for all candidates.

   Digest schema per §7.4:
       candidate_id, analyzes_support_count, analyzes_support_items,
       affects_support_present, affects_support_items,
       observationally_ungrounded, causal_grounding_absent,
       degraded_classification_count, confidence_cap


.. py:function:: build_epistemics_run_summary(causality_candidates, results, evidence_bundle, calibration_profile_name = None, calibration_profile_version = None)

   Build run_manifest.epistemics_summary (§7.4 manifest consequence).

   Records: hit counts by epistemic class, supersession edges, degraded
   classification counts by artifact type, classification resolution level
   distribution, and calibration profile reference.